GEMS

secondsPaperCodeNeed >5GB

GNN-based protein-ligand binding affinity prediction with language model embeddings

Readiness

Readiness: Need >5GB
Smoke: not_tested
Artifacts: not_applicable
Production: need_gt5gb_cuda_pyg_24gb_vram_runtime_pending
Storage: 5-20GB
Strategy: shared CUDA PyTorch/PyG GEMS SIF + trained model/precomputed example overlay
Next: Build/reuse CUDA 11.7 PyTorch 2.0/PyG GEMS SIF, stage trained models plus tiny precomputed example, run direct Slurm GPU smoke only on adequate VRAM, then wire runner.

No obvious large model/database dependency identified; validate upstream docs before install. 2026-07-05 SOP batch recheck: backend inventory found no staged GEMS source/models/precomputed examples/SIF/artifacts/runner. Upstream requires CUDA/PyTorch/PyG and recommends RTX3090-class 24GB VRAM; preprocessing can need about 100GB unless precomputed data is used. Reclassified from Smoke Pending to Need >5GB.